Mode Filter Research Articles

BiLSTM-Filt: Neural network for radar word segmentation

Radar word extraction is the analysis foundation for multi-function radars (MFRs) in electronic intelligence (ELINT). Although neural networks enhance performance in radar word extraction, current research still faces challenges from complex electromagnetic environments and unknown radar words. Therefore, in this paper, we propose a promising two-stage radar word extraction framework, consisting of segmentation and recognition. To fill the vacancy of radar word segmentation, we establish the mathematical model from the time series analysis viewpoint and design a novel segmentation neural network based on Bi-direction Long Short-Term Memory with a filter module (BiLSTM-Filt). Specific radar word structure characteristics are extracted by training the network and applied for detecting radar words in the pulse train. To further improve segmentation performance, a bounding box regression method is designed to merge information from sub-region structures. Simulation experiments on a typical MFR, Mercury, reveal that the proposed method can outperform the baseline methods within complex electromagnetic environments, containing corrupted environments, various pulse backgrounds, and variable pulse train lengths. Due to the artificial design structure, the proposed method can also make a trial on unknown radar word segmentation.

Multi-feature fusion and multi-attention deep network for enhancing road extraction in remote sensing images

ABSTRACT Road detection in remote sensing (RS) images plays a critical role in applications ranging from urban planning to autonomous navigation systems. However, accurate road extraction remains a challenging task due to the presence of textual-similar objects that can be visually confused with roads, and shadows that can obscure road features. For this regard, we present a novel end-to-end network that leverages multi-feature fusion and multi-attention. Multi-level dual residual blocks are introduced to capture multi-scale and multi-level road features. A channel attention feature fusion module fuses road features from the decoder while suppressing coarse-grain noises. Moreover, a difference channel feature fusion module filters out the interfering texture-similar features and irrelevant background features. Subsequently, a spatial channel feature enhancement module is designed to identify the mis-segmented regions and reclassify these pixels. We conducted extensive experiments on diverse datasets, demonstrating the effectiveness of our approach in improving road detection accuracy, especially in areas with shadows and textual-similar objects. The results indicate that our approach outperforms state-of-the-art methods in handling textual-similar objects and shadow-occluded roads, making it a valuable contribution to the field of road extraction in remote sensing images.

"Seeing" ENF From Neuromorphic Events: Modeling and Robust Estimation.

Most artificial lights exhibit subtle fluctuations in intensity and frequency in response to the influence of the grid's alternating current, providing the potential to estimate the Electric Network Frequency (ENF) from conventional frame-based videos. Nevertheless, the performance of Video-based ENF (V-ENF) estimation largely relies on the imaging quality and thus may suffer from significant interference caused by non-ideal sampling, scene diversity, motion interference, and extreme lighting conditions. In this paper, we show that the ENF can be extracted without the above limitations from a new modality provided by the so-called event camera, a neuromorphic sensor that encodes the light intensity variations and asynchronously emits events with extremely high temporal resolution and high dynamic range. Specifically, we formulate and validate the physical mechanism for the ENF captured in events and then propose a simple yet robust Event-based ENF (E-ENF) estimation method through mode filtering and harmonic enhancement. To validate the effectiveness, we build the first Event-Video ENF Dataset (EV-ENFD) and its extension EV-ENFD+ with diverse scenarios, including static, dynamic, and extreme lighting scenes. Comprehensive experiments have been conducted on our proposed datasets, showcasing that our proposed E-ENF significantly outperforms the V-ENF in extracting accurate ENF traces, especially in challenging environments.

SF-GPT: A training-free method to enhance capabilities for knowledge graph construction in LLMs

Knowledge graphs (KGs) are constructed by extracting knowledge triples from text and fusing knowledge, enhancing information retrieval efficiency. Current methods for knowledge triple extraction include ”Pretrain and Fine-tuning” and Large Language Models (LLMs). The former shifts effort from manual extraction to dataset annotation and suffers from performance degradation with different test and training set distributions. LLMs-based methods face errors and incompleteness in extraction. We introduce SF-GPT, a training-free method to address these issues. Firstly, we propose the Entity Extraction Filter (EEF) module to filter triple generation results, addressing evaluation and cleansing challenges. Secondly, we introduce a training-free Entity Alignment Module based on Entity Alias Generation (EAG), tackling semantic richness and interpretability issues in LLM-based knowledge fusion. Finally, our Self-Fusion Subgraph strategy uses multi-response self-fusion and a common entity list to filter triple results, reducing noise from LLMs’ multi-responses. In experiments, SF-GPT showed a 55.5% increase in recall and a 32.6% increase in F1 score on the BDNC dataset compared to the UniRel model trained on the NYT dataset and achieved a 5% improvement in F1 score compared to GPT-4+EEF baseline on the WebNLG dataset in the case of a fusion round of three. SF-GPT offers a promising way to extract knowledge from unstructured information.

FilterGNN: Image feature matching with cascaded outlier filters and linear attention

The cross-view matching of local image features is a fundamental task in visual localization and 3D reconstruction. This study proposes FilterGNN, a transformer-based graph neural network (GNN), aiming to improve the matching efficiency and accuracy of visual descriptors. Based on high matching sparseness and coarse-to-fine covisible area detection, FilterGNN utilizes cascaded optimal graph-matching filter modules to dynamically reject outlier matches. Moreover, we successfully adapted linear attention in FilterGNN with post-instance normalization support, which significantly reduces the complexity of complete graph learning from O(N2) to O(N). Experiments show that FilterGNN requires only 6% of the time cost and 33.3% of the memory cost compared with SuperGlue under a large-scale input size and achieves a competitive performance in various tasks, such as pose estimation, visual localization, and sparse 3D reconstruction.

Efficient time series adaptive representation learning via Dynamic Routing Sparse Attention

Time series prediction plays a crucial role in various fields but also faces significant challenges. Converting original 1D time series data into 2D data through dimension transformation allows capturing more hidden features but incurs high memory consumption and low time efficiency. We have designed a sparse attention mechanism with dynamic routing perception called Dynamic Routing Sparse Attention (DRSA) to address these issues. Specifically, DRSA can effectively handle variations of complex time series data. Meanwhile, under memory constraints, the Dynamic Routing Filter (DRF) module further refines it by filtering the blocked 2D time series data to identify the most relevant feature vectors in the local context. We conducted predictive experiments on six real-world time series datasets with fine granularity and long sequence dependencies. Compared to eight state-of-the-art (SOTA) models, DRSA demonstrated relative improvements ranging from 4.18% to 81.02%. Furthermore, its time efficiency is 2 to 5 times higher than the baseline. Our code and dataset will be available at https://github.com/wwy8/DRSA_main.

Vehicle trajectory prediction based on cross-attention and multilevel spatio-temporal features

Accurately predicting vehicle trajectories is crucial for motion planning in autonomous driving. Some existing trajectory prediction models employ Long Short-Term Memory (LSTM) networks to encode trajectory information, but this approach may lead to information loss. Additionally, the sparsity of the grid representation may limit the model’s ability to extract vehicle interaction features. To address these issues, this paper proposes a vehicle trajectory prediction model based on cross-attention and multilevel spatio-temporal features. The proposed model incorporates a spatio-temporal feature extraction module designed to capture the dynamic evolution of vehicle states across both time and space. Furthermore, it employs a temporal cross-attention module that reuses hidden states to select spatio-temporal features closely related to the target vehicle’s historical state, thus compensating for the temporal information loss during the encoding process. Additionally, a spatial cross-attention module, informed by social context, is deployed to analyze the dynamic interactions among vehicles. This module filters spatio-temporal interaction features with a significant impact on the target vehicle’s trajectory. The model is trained and tested using the NGSIM dataset and the highD dataset. Compared with other models, the model presented in this paper shows higher prediction accuracy in long-term trajectory prediction.

Scale reduction Transformer-based soft measurement of oil–water two-phase flow

In addressing the challenge of parameter measurement in high-water-content oil–water two-phase flow, we independently develop a dual-helix microwave sensor measurement system. Using this system, we conduct vertically upward experiments on oil–water two-phase flow and collect time-series signals characterizing microwave attenuation. Based on signal characteristics, we design a scale reduction Transformer soft measurement model (SRSM-Transformer). The time filter module effectively facilitates the aggregation and embedding of channel features with diverse physical significance. Additionally, the scale reduction temporal module models global information from a multi-resolution perspective, enabling the model to flexibly extract multi-scale feature information from fluid signals. Demonstrating pronounced superiority over existing models, our proposed model achieves recognition accuracies of 93.63% for total flow rate and 95.73% for water content. This research provides a new direction for modeling complex industrial oil–water two-phase flow systems and measuring multi-coupled key parameters.

Deep learning approach for skin melanoma and benign classification using empirical wavelet decomposition.

Melanoma is a malignant skin cancer that causes high mortality. Early detection of melanoma can save patients' lives. The features of the skin lesion images can be extracted using computer techniques to differentiate early between melanoma and benign skin lesions. A new model of empirical wavelet decomposition (EWD) based on tan hyperbolic modulated filter banks (THMFBs) (EWD-THMFBs) was used to obtain the features of skin lesion images by MATLAB software. The EWD-THMFBs model was compared with the empirical short-time Fourier decomposition method based on THMFBs (ESTFD-THMFBs) and the empirical Fourier decomposition method based on THMFBs (EFD-THMFBs). The accuracy rates obtained for EWD-THMFBs, ESTFD-THMFBs, and EFD-THMFBs models were 100%, 98.89%, and 83.33%, respectively. The area under the curve (AUC) was 1, 0.97, and 0.91, respectively. The EWD-THMFBs model performed best in extracting features from skin lesion images. This model can be programmed on a mobile to detect skin lesions in rural areas by a nurse before consulting a dermatologist.

The Blossoming of Ultrasonic Metatransducers.

Key requirements to boost the applicability of ultrasonic systems for in situ, real-time operations are low hardware complexity and low power consumption. These features are not available in present-day systems due to the fact that US inspections are typically achieved through phased arrays featuring a large number of individually controlled piezoelectric transducers and generating huge quantities of data. To minimize the energy and computational requirements, novel devices that feature enhanced functionalities beyond the mere conversion (i.e., metatransducers) can be conceived. This article reviews the potential of recent research breakthroughs in the transducer technology, which allow them to efficiently perform tasks, such as focusing, energy harvesting, beamforming, data communication, or mode filtering, and discusses the challenges for the widespread adoption of these solutions.

A Plug-In Graph Neural Network to Boost Temporal Sensitivity in fMRI Analysis.

Learning-based methods offer performance leaps over traditional methods in classification analysis of high-dimensional functional MRI (fMRI) data. In this domain, deep-learning models that analyze functional connectivity (FC) features among brain regions have been particularly promising. However, many existing models receive as input temporally static FC features that summarize inter-regional interactions across an entire scan, reducing the temporal sensitivity of classifiers by limiting their ability to leverage information on dynamic FC features of brain activity. To improve the performance of baseline classification models without compromising efficiency, here we propose a novel plug-in based on a graph neural network, GraphCorr, to provide enhanced input features to baseline models. The proposed plug-in computes a set of latent FC features with enhanced temporal information while maintaining comparable dimensionality to static features. Taking brain regions as nodes and blood-oxygen-level-dependent (BOLD) signals as node inputs, GraphCorr leverages a node embedder module based on a transformer encoder to capture dynamic latent representations of BOLD signals. GraphCorr also leverages a lag filter module to account for delayed interactions across nodes by learning correlational features of windowed BOLD signals across time delays. These two feature groups are then fused via a message passing algorithm executed on the formulated graph. Comprehensive demonstrations on three public datasets indicate improved classification performance for several state-of-the-art graph and convolutional baseline models when they are augmented with GraphCorr.

An adaptive feature mode decomposition-guided phase space feature extraction method for rolling bearing fault diagnosis

Abstract The extraction of fault features from rolling bearings is a challenging and highly important task. Since they have complex operating conditions and are usually under a strong noise background. In this study, a novel approach termed phase space feature extraction guided by an adaptive feature mode decomposition (AFMDPSFE) is proposed to detect subtle faults in rolling bearings. Initially, a new method using Kullback-Leiber divergence is introduced to automatically select the optimal mode number and filter length for the decomposition of vibration signals, facilitating the automatic extraction of optimal components and ensuring efficient screening. This eliminates the need for manual configuration of feature mode decomposition parameters. Furthermore, a criterion that could determine two crucial parameters to capture system dynamics characteristics in phase space reconstruction is embedded into AFMDPSFE algorithm. Subsequently, a series of high-dimensional independent components is derived. The envelope spectrum of the principal component exhibiting the highest kurtosis value is computed to achieve fault identification, consequently enhancing the separation of signal from noise. Both simulations and experimental results confirm the effectiveness of AFMDPSFE approach. A comparison analysis shows the excellent performance of AFMDPSFE in extracting fault features from significant noise interference.

A 0.00179 mm2/Ch Chopper-Stabilized TDMA Neural Recording System With Dynamic EOV Cancellation and Predictive Mixed-Signal Impedance Boosting.

This article presents a digitally-assisted multi-channel neural recording system. The system uses a 16-channel chopper-stabilized Time Division Multiple Access (TDMA) scheme to record multiplexed neural signals into a single shared analog front end (AFE). The choppers reduce the total integrated noise across the modulated spectrum by 2.4 × and 4.3 × in Local Field Potential (LFP) and Action Potential (AP) bands, respectively. In addition, a novel impedance booster based on Sign-Sign least mean squares (LMS) adaptive filter (AF) predicts the input signal and pre-charges the AC-coupling capacitors. The impedance booster module increases the AFE input impedance by a factor of 39 × with a 7.1% increase in area. The proposed system obviates the need for on-chip digital demodulation, filtering, and remodulation normally required to extract Electrode Offset Voltages (EOV) from multiplexed neural signals, thereby achieving 3.6 × and 2.8 × savings in both area and power, respectively, in the EOV filter module. The Sign-Sign LMS AF is reused to determine the system loop gain, which relaxes the feedback DAC accuracy requirements and saves 10.1 × in power compared to conventional oversampled DAC truncation-error ΔΣ-modulator. The proposed SoC is designed and fabricated in 65 nm CMOS, and each channel occupies 0.00179 mm2 of active area. Each channel consumes 5.11 μW of power while achieving 2.19 μVrms and 2.4 μVrms of input referred noise (IRN) over AP and LFP bands. The resulting AP band noise efficiency factor (NEF) is 1.8. The proposed system is verified with acute in-vivo recordings in a Sprague-Dawley rat using parylene C based thin-film platinum nanorod microelectrodes.

A frequency selection network for medical image segmentation

Existing medical image segmentation methods may only consider feature extraction and information processing in spatial domain, or lack the design of interaction between frequency information and spatial information, or ignore the semantic gaps between shallow and deep features, and lead to inaccurate segmentation results. Therefore, in this paper, we propose a novel frequency selection segmentation network (FSSN), which achieves more accurate lesion segmentation by fusing local spatial features and global frequency information, better design of feature interactions, and suppressing low correlation frequency components for mitigating semantic gaps. Firstly, we propose a global-local feature aggregation module (GLAM) to simultaneously capture multi-scale local features in the spatial domain and exploits global frequency information in the frequency domain, and achieves complementary fusion of local details features and global frequency information. Secondly, we propose a feature filter module (FFM) to mitigate semantic gaps when we conduct cross-level features fusion, and makes FSSN discriminatively determine which frequency information should be preserved for accurate lesion segmentation. Finally, in order to make better use of local information, especially the boundary of lesion region, we employ deformable convolution (DC) to extract pertinent features in the local range, and makes our FSSN can focus on relevant image contents better. Extensive experiments on two public benchmark datasets show that compared with representative medical image segmentation methods, our FSSN can obtain more accurate lesion segmentation results in terms of both objective evaluation indicators and subjective visual effects with fewer parameters and lower computational complexity.

OEDG: Oscillation-eliminating discontinuous Galerkin method for hyperbolic conservation laws

Suppressing spurious oscillations is crucial for designing reliable high-order numerical schemes for hyperbolic conservation laws, yet it has been a challenge actively investigated over the past several decades. This paper proposes a novel, robust, and efficient oscillation-eliminating discontinuous Galerkin (OEDG) method on general meshes, motivated by the damping technique (see J. Lu, Y. Liu, and C. W. Shu [SIAM J. Numer. Anal. 59 (2021), pp. 1299–1324]). The OEDG method incorporates an oscillation-eliminating (OE) procedure after each Runge–Kutta stage, and it is devised by alternately evolving the conventional semidiscrete discontinuous Galerkin (DG) scheme and a damping equation. A novel damping operator is carefully designed to possess both scale-invariant and evolution-invariant properties. We rigorously prove the optimal error estimates of the fully discrete OEDG method for smooth solutions of linear scalar conservation laws. This might be the first generic fully discrete error estimate for nonlinear DG schemes with an automatic oscillation control mechanism. The OEDG method exhibits many notable advantages. It effectively eliminates spurious oscillations for challenging problems spanning various scales and wave speeds, without necessitating problem-specific parameters for all the tested cases. It also obviates the need for characteristic decomposition in hyperbolic systems. Furthermore, it retains the key properties of the conventional DG method, such as local conservation, optimal convergence rates, and superconvergence. Moreover, the OEDG method maintains stability under the normal Courant–Friedrichs–Lewy (CFL) condition, even in the presence of strong shocks associated with highly stiff damping terms. The OE procedure is nonintrusive, facilitating seamless integration into existing DG codes as an independent module. Its implementation is straightforward and efficient, involving only simple multiplications of modal coefficients by scalars. The OEDG approach provides new insights into the damping mechanism for oscillation control. It reveals the role of the damping operator as a modal filter, establishing close relations between the damping technique and spectral viscosity techniques. Extensive numerical results validate the theoretical analysis and confirm the effectiveness and advantages of the OEDG method.

Compact high extinction ratio high-order mode pass filter based on inverse-designed ultra-compact unidirectional mode converter

Mode division multiplexing (MDM) technology provides a pathway to enhance channel capacity beyond wavelength division multiplexing, positioning it as a pivotal advancement for next generation optical communications. Mode filters are essential for the low-loss transmission of specific modes and the reduction of modal crosstalk, thereby enhancing the feasibility of MDM systems. Although suppressing high-order mode is relatively straightforward, effectively blocking low-order modes poses a more intricate challenge. In this paper, we introduce a high-order mode pass strategy, effectively blocking low-order modes using the unidirectional mode converters. Specifically, a TE1 high-order mode pass filter (HOMPF) is demonstrated on a silicon-on-insulator platform, utilizing a unique inverse-designed ultra-compact unidirectional TE0-TE1 mode converter. Experimental results show the TE1-TE1 insertion loss of the HOMPF of below 1.0 dB and an average TE0-TE0 extinction ratio of 36.8 dB (42.1 dB for 2-cascaded HOMPF) within the C-band range of 1525-1565 nm. Additionally, the scalability of the HOMPF structure is explored, with simulations demonstrating a TE2 HOMPF. The proposed HOMPFs feature simplicity, compactness, low loss, and high extinction ratio, making them promising components for mode manipulation in MDM systems.

Simulation of Modal Control of Metal Mode-Filtered Vertical-Cavity Surface-Emitting Laser.

In this study, a novel metal-dielectric film mode filter structure that can flexibly regulate the transverse mode inside vertical-cavity surface-emitting lasers (VCSELs) is proposed. The number, volume, and stability of transverse modes inside the VCSEL can be adjusted according to three key parameters-the oxide aperture, the metal aperture, and the distance between the oxide aperture and the metal aperture-to form a flexible window, and a new parameter is defined to describe the mode identification. This study provides a complete simulation theory basis and calculation method, which is of great significance for the optical mode control in VCSELs.

Estimation of influence of the substrate material of the modal filter with electromagnetic absorber on the difference of the modal delays

Estimation of influence of the substrate material of the modal filter with electromagnetic absorber on the difference of the modal delays

Monocular Depth Estimation via Self-Supervised Self-Distillation.

Self-supervised monocular depth estimation can exhibit excellent performance in static environments due to the multi-view consistency assumption during the training process. However, it is hard to maintain depth consistency in dynamic scenes when considering the occlusion problem caused by moving objects. For this reason, we propose a method of self-supervised self-distillation for monocular depth estimation (SS-MDE) in dynamic scenes, where a deep network with a multi-scale decoder and a lightweight pose network are designed to predict depth in a self-supervised manner via the disparity, motion information, and the association between two adjacent frames in the image sequence. Meanwhile, in order to improve the depth estimation accuracy of static areas, the pseudo-depth images generated by the LeReS network are used to provide the pseudo-supervision information, enhancing the effect of depth refinement in static areas. Furthermore, a forgetting factor is leveraged to alleviate the dependency on the pseudo-supervision. In addition, a teacher model is introduced to generate depth prior information, and a multi-view mask filter module is designed to implement feature extraction and noise filtering. This can enable the student model to better learn the deep structure of dynamic scenes, enhancing the generalization and robustness of the entire model in a self-distillation manner. Finally, on four public data datasets, the performance of the proposed SS-MDE method outperformed several state-of-the-art monocular depth estimation techniques, achieving an accuracy (δ1) of 89% while minimizing the error (AbsRel) by 0.102 in NYU-Depth V2 and achieving an accuracy (δ1) of 87% while minimizing the error (AbsRel) by 0.111 in KITTI.

Developing self-calibrating system for fiber Bragg grating based guided wave sensing under changing temperature conditions

Fiber Bragg grating (FBG) sensors have long been thought of as the ideal sensors for structural health monitoring (SHM) due to their small size, light weight, ability to be embedded and ability to be multiplexed. So, FBG sensors have been commonly used for strain based SHM. In recent times, a renewed interest is seen in the use of FBG sensors for guided wave (GW) measurements using the edge filtering approach which increases the sensitivity several folds. They offer several unique opportunities for GW based SHM such as allowing mode filtering, acoustic coupling, etc. Unfortunately, more wide spread research is limited by the steep learning curve. Also, the use of FBG in real applications is still in its infancy due to the need of calibration of the system when the ambient temperature conditions change. This paper precisely tries to address these two shortcomings. For overcoming the steep learning curve, a detailed discussion on the hardware for the FBG based GW sensing is provided. Following the discussion a step-by-step approach is outlined for incorporating the sensors. A detailed trouble-shooting guide is developed based on the immense experience of the authors in this field. This exercise will allow easier adoption of the technique and stimulate more research in the topic. The exercise also allows us to highlight the safeguards and the features that need to be included in the system which will be self-calibrating. Once the design parameters are established a self-calibrating autonomous FBG based sensing system is developed. The developed system is tested in ambient conditions over an extended period in the day capturing the ambient temperature changes. The system is also tested in a larger temperature range (25 ∘C–65 ∘C). The results indicate that indeed the self-calibrating system works effectively. Some sensitivity studies to determine the performance in terms of system reaction time have also been provided. Such a ‘smart’ autonomous system for GW sensing has not been presented to the best of the author’s knowledge and is the key novelty of the presented work. Furthermore, the detailed discussions and troubleshooting guide will help introduce more people to this field of study which will lead to more radical development of the field.